Agents of Change: Walk Into the Room Knowing Everyone — AI-Powered Event Prep

My name is Janette Roush, and I’m the SVP of Innovation and the Chief AI Officer for Brand USA, and you are joining our Agents of Change webinar series. Today the topic is: how can you walk into the room knowing everyone? How can AI assist us with preparation for a business event like IMEX, coming up next week? We just finished Brand USA’s Travel Week event. So how can we use artificial intelligence to prepare us for those events before we actually get on the floor?

To give you an example of what that output is going to look like, I saw that Sally has logged in, so thank you, Sally, for joining. I use AI, specifically Claude, to create a face sheet for me before I go to a conference. And I actually made a face sheet of everybody who had registered for today’s Zoom. I am able to do that because of all of the connectors we have set up connecting Claude, which is our primary AI tool at Brand USA, to things like our Outlook, Salesforce, which is our CRM, and a system of note-taking that I have set up on my own computer to capture information from the meetings I have all year long. So to prepare for something like today’s webinar, I can go to Claude and say, “Hey, would you build me a face sheet that includes information on each of the people who have registered for today’s webinar, so that I show up for the webinar a little bit more prepared?”

Thank you, Sally, for your generosity in letting me share your information here. I’m going to show you how you can do this to walk into IMEX next week, or into any event you might be attending in the future.

We know what these shows look like. This could very well be what your appointment schedule looks like for IMEX on Tuesday. It is just back-to-back meetings. It’s the same thing with our Travel Week event. And what that means is by the time you get to the end of the week, it is really hard to remember who said what. It is just a pile of conference badges in your memory. And if you’re taking notes, perhaps those notes are scribbled in the margins of pieces of paper. I know when I was taking appointments at shows like IPW on behalf of selling Broadway tickets, coming back to the office and trying to decipher what I had scrawled down was always a challenge. I did not have a great system for it.

And we know when we attend these conferences, that’s actually our time to connect in person. The whole point of having AI available to us is, in theory, that it makes us more available for the in-person things that we truly value. So it feels like a waste if we actually have the opportunity to do an in-person meeting and we spend all of that time figuring out how to write notes, rather than connecting with the human being sitting across from us.

And I know this isn’t just my own lived experience. Salesforce did a study about this last year, around August or September, where they interviewed people around the world who work in sales to find out how much of their work week they spend on non-selling work. Even the manual data entry piece of that work is 13% of your work week. So it’s a real problem. When we have the opportunity to be selling, we want to spend that time selling. And AI can now do the part of the work that you don’t want to do. That’s why we have it. It’s not to take away the things we love about our jobs. It’s to take away those rote tasks so that you can be present for the 15 minutes that you have in the chair.

The trick, though, about working with AI or any tool like this is that it’s not going to remember information that you haven’t captured. It’s just like our brains, which are able to forget things, or like working with your CRM. If you don’t put something into Simpleview or HubSpot or Salesforce or Tempest, you can’t get that information back out again unless you happen to be in the right place at the right time to remember it. So part of this process is showing how you capture all of the information that is shared in a meeting, so that it actually becomes useful to you as part of the follow-up process, and then prepares you for the next time you meet with someone.

Step one is that you are going to want to record the meeting, and I’m going to show you a few different ways that you can do that. The device you see in the image here is a physical recording device. It is called Plaud, and that is a little lapel pin. It comes with a variety of attachments, so you can wear it on your wrist, or you can wear it as a pendant, or use a little clip or magnet to attach it to your clothes. You press the button in the middle, and it will record what is being said in your conversation. It doesn’t actually retain the recording. What it does is create a transcript of that recording. It sends that to an app on your phone through a Bluetooth connection, and then the app on your phone is able to push it to the cloud so that the AI tool of your choosing is able to access that transcript.

Another way that you can capture meeting notes is using Granola, and that’s what we use at Brand USA. Granola works for Zoom video calls, but it also works as an app on your phone to record an in-person meeting at a place like Brand USA Travel Week or IMEX. The way Granola works specifically is that if you are hosting a meeting, there’s not a bot that joins the meeting. You might be familiar with some AI note-takers like Read.ai, Otter, or Fireflies, which actually open up a separate window on a Zoom call and join the call. Granola works in the background, and it captures the audio of whatever your computer can hear, which is good and bad. You don’t have the clunky extra item joining your call, but that puts the onus on you as the user to disclose that you are recording, because there is no external signal to anybody that a transcript is being made of the call.

It’s part of Brand USA’s AI policy that we want to disclose AI when we’re using it, because people don’t like to feel like you’re tricking them or sneaking around. So you would want to disclose your use of something like Granola. On the flip side, we should probably now all assume that every time we are on a call, it is being recorded, possibly without our knowledge or consent, and we should judge what we say and disclose on those calls accordingly. Granola makes a transcript of everything that is said. It is also an app that opens on your computer, so you can type in notes at the same time, and it will save all of those notes. You are then able to connect Granola on the back end to the AI tools that you use.

Whether or not that’s possible is going to depend both on your IT team and what permissions they have granted your organization, and on what specific AI tools you are using. Both Claude and ChatGPT have a large number of connectors built into their systems, so it’s really easy to connect something like Granola to Claude. Granola has a free version, but that’s going to give you just the notes from a call. If you want to see the full transcript of the call, that’s when you’re going to want the paid plan, which starts at fourteen dollars a month.

So what should you use to record? You can use your phone with Granola or a similar app open on it. The voice memo, which both Android phones and iPhones have built in, is how you can use your phone to make a transcript without buying any external services. If you’re using a meeting app like Granola, something useful is that if it’s connected to your Outlook or Google account, it has access to your calendar. So it knows, based on the time, exactly what meeting you are in and what people you are talking to, assuming they are on the calendar invite. That can be really convenient if you are able to match up your IMEX schedule with the actual times that you were talking to people, because then you don’t have to go back and figure it out after the fact.

A dedicated recorder like Plaud could possibly provide better audio. There’s also the ability inside of Plaud to assign speakers, so it’ll recognize who has which speaking voice. Across different transcripts in Plaud, it will know when I’m talking because it knows my voice.

To make sure the transcripts you capture at one of these events are useful, you need to make sure at the start that you are asking for permission to record. Obviously, if somebody says they’re not comfortable with it, you don’t want to do it. I will say that with folks testing this out on our side, both at Travel Week and at IPW before that, the acceptance of using a transcription tool is very high. People now know and accept that this is happening. You want to say at the beginning of your recording who you are, who the people around the table are, and what the company is, because if it doesn’t match up to a calendar invite, that keeps it clear when you’re reviewing the transcript later exactly what meeting it was. And it’s good meeting hygiene anyway to end any meeting by discussing what commitments you made, what the next steps are, and whether there are due dates for those next steps.

The meeting transcription tool will read through your notes and come up with those next steps. But why not make it easier on the tool and actually say to the person you are meeting with what those next steps are? That’s just a good idea at meetings anyway.

If you’re not familiar with connectors, this is what the connector screen looks like in Claude. You would navigate to the settings of your preferred AI tool. In Claude, there is a section called Connectors, and you can click through to add the permissions for that. In your corporate account, it’s not going to be automatically set up on day one. That is something your IT team needs to provision for you. And if you go back to the initial prompt I shared a few slides ago about Marcus Donnelly, the imaginary meeting planner — what did I discuss with him the last time I saw him, and what did I promise him? — if you’ve been recording these meetings over time, you’re going to start to have actual answers that you can deliver back to this person using Claude.

Depending on your setup at your organization, there are a number of different ways that you can make this happen. While the examples I show are generally going to use Claude, you don’t have to be in Claude to benefit from the system I’m walking through. If you are a Claude user and you’re using Granola, there’s a connector that will help you read the transcripts of your meetings. You can use Cowork inside of Claude to write the files I’m going to show you how to write. If you want to connect it to your Outlook or Gmail, there’s a native connector. If you want it to connect to your CRM, that’s going to be contingent on what CRM you use. Some CRMs have connectors available. I’ve not used HubSpot, but they are very pro-AI, and they are making sure that everything you do in HubSpot can happen either logged into HubSpot or inside of your preferred AI tool.

That’s been my experience working in Salesforce as well. If your current CRM doesn’t offer this, it is time for those companies to start building in that functionality, and that’s pressure the industry could start putting on those companies. If you’re a ChatGPT user, it’s going to be the same thing. On a paid plan, you’ll be able to connect to Granola automatically. So you don’t have to copy and paste a transcript from Granola into ChatGPT. You can just say inside of ChatGPT, “Hey, look up that meeting I had last week,” and it can do that through this connector. If you want ChatGPT to read and write files for you, that’s called Cowork inside of Claude, and it is called Work inside of ChatGPT.

ChatGPT has the same CRM, Outlook, and Gmail connectors that Claude has. Now, if you’re a user of Copilot at your organization, it’s not going to have the same power in terms of connectors. This is when you’re going to be copying the transcript and pasting it into Copilot. I am not familiar with OneNote notebooks, but if you are a user of Copilot, that is the way Copilot is able to read and write files for you, so that is the avenue you’re going to want to explore for what I’m teaching today. Reading your email is going to be the clearest answer in Copilot, because it’s made by Microsoft. If you’re using Copilot, you’re likely not a Gmail user or in Google Workspace. You’re already in the Microsoft ecosystem, so those connectors are built in.

For a CRM, you’re going to be exporting reports. If you’re working with a CRM that doesn’t have a connector available, exporting the report is also what you would do. And if you don’t have any connectors at all, everything is going to be copy and paste, but you can still do this.

What I’m going to show you now is how you can create a connected web of all of your contacts. The first step is making a folder on your computer that is easily accessible to your AI tool. You can name this whatever you want. For this example, I’m calling it My Notes. The great thing about ChatGPT and Claude is that you don’t have to actually make this folder yourself, because those tools are able to read and write documents on your computer. You can say, “Hey, would you create a folder called My Notes in My Documents,” or on your desktop, or wherever you want to save this folder, “and then put these four specific subfolders in the folder?” Honestly, you can take a screenshot of this slide and paste it into ChatGPT or Claude, and it will do it without you needing to retype anything.

You could also, if you are making a transcript of this webinar, feed that transcript to your AI tool and say, “Hey, I want to set up the thing that she talked about. Can you do that?” That is how I set up this system originally. I didn’t invent the system. Andrej Karpathy created something that he calls Second Brain, which uses this format. He is an AI researcher who I believe left OpenAI last year and is now at Anthropic. He’s a cutting-edge thought leader and researcher in this field, and this is how he works. I didn’t reinvent the wheel. I just pasted a link to a YouTube video into my Claude account and said, “Show me what I need to do in order to do what he does.” So copy that tip.

If you’re using Claude, you’re going to click the Cowork button, which allows Claude to read and write folders on your computer. Then you’re going to select the My Notes folder. You are now connecting your Claude conversation to that specific folder. If you’re in ChatGPT, you’re going to click on Work, and it’ll be the same thing. You’re going to select the folder after that.

So what goes in these folders? Honestly, you barely need to know this part, because Claude is going to do the work for you. You have four folders. One folder is named Meetings. It’s going to hold the transcript or call note from every meeting you have, including what they asked for and what you promised you would do for them. The second folder is called People. There’s going to be one file for each person you meet with inside of that folder. You’re going to do the same thing for companies. Every single company gets its own file. If that sounds like a lot of work, don’t worry about it, because Claude writes these files for you. You are not the one physically writing and updating these files.

The fourth piece is topics. What did you talk about? What meetings were you planning? What dates was the partner looking at? You can create a hub for each theme you are discussing. This is what it would look like once it’s filled in. In Meetings, you have a list of meetings or phone calls with the dates. In People, we have a list of names. In Companies, we have a list of the companies. And in Topics, here’s convention sales, here’s everybody who’s looking for walkable districts for their meetings, and here are all of the conversations that tie back to room blocks. What makes this really powerful is that all of these things are linked together. So for Marcus Donnelly here, who works at Hartwell Pierce, there’s a wiki link inside the folder that Claude will write for you.

That way, you can go to that company page, or ask Claude to go to the company page. I will tell you, I have hundreds of these files at this point, and I don’t actually go and look at any of them. It’s what Claude uses as a reference to build tools for me. But it can look at one of these company pages, and there are links to every single person who works at that company. So it instantly knows additional context to prepare you for a follow-up meeting with somebody from that company. Both the people file and the company file are also going to link to any meetings you had with those people. So you can see how you start to make connections that might have been more difficult to make before. And for topics, I think that’s really helpful.

Certainly in my work, if I’m keeping track of different projects, I can ask Claude something about Travel Week Canada, and it can look to see if that topic has been discussed at any meetings. It can immediately connect me to the meetings where I discussed Brand USA Travel Week Canada, the people who were in those meetings, and the companies that were involved. It’s creating that web of context and information, and that is what makes AI powerful. Without context, AI really is limited to serving as a souped-up Google search for you, or as the tool that rewrites your emails so you don’t come off as rude. Those are great use cases, but it’s possible to do a lot more with AI.

Now, this is the prompt you would want to give your AI tool to start setting this up. Let’s assume you have already pointed your AI tool of choice at a folder called My Notes, that there are four subfolders in that folder, and that you also have a connector set up for Granola. You can say, “Hey, please look at my Granola meetings for today. I want you to write one note per meeting, and I want you to create a file for every single person, company, and topic that was part of the meeting. I want you to spell out who was there, what they asked for, what I promised them, and what the next steps are for this conversation, and then make sure that you are linking these things together.”

The double-bracketed link you see here is used in a type of file called Markdown. Markdown looks like plain text with a little bit of formatting on it. A human can still read it, but it’s very easy for AI to read and to understand the hierarchy of the information it’s looking at. It sees those double brackets as a wiki link to some other piece of information in its files. I also like to save the entire transcript of a meeting rather than just a summary, because summaries can sometimes leave out a lot of helpful information. If you have the space to save that much information, it’s helpful to have the entire transcript saved in this folder system.

Once you have this prompt set up, you can say, “Hey, pull today’s Granola meetings and file them,” and “file them” refers to the process we just walked through. If you are using either Claude or ChatGPT, you are able to take a prompt like this and save it. In Claude, that’s called a skill, which is just a reusable prompt that you can invoke by saying, “Hey, file my Granola meetings.” Because it’s not looking for some special magic language and it operates in plain English, it will know what that means, and it will execute the prompt you have set up. All AI tools have this. In Gemini, which I’m not really going into in this presentation, it’s called a Gem. In ChatGPT, it was called a custom GPT, and I think they’ve made changes to that, so I would have to research it if you had a question on how to do that. But all AI systems have this kind of repeatable prompt.

As I said before, every single note is going to link three different ways. Every person is going to get a profile that remembers your history together, because every time you have another meeting with that person, Claude isn’t going to write a new file for them from scratch. It’s going to look at the Markdown file it made before and amend it to add the new information. It’s the same thing with companies and topics. This is how the system gets richer over time. You can see that Marcus is here, with our full relationship history spelled out. The text you see there is Markdown. Again, it’s easy to read. The two pound signs before “Relationship History” just indicate that it’s a header, and that helps Claude understand that this information sits above the rest of the information. It shows the hierarchy of information to the AI tool, but we can still read it. And the ultimate outcome is that all of these things are connected together.

So how do we take this to the next step, so that you walk into the convention center next week and actually feel like you know the people who are in the room? This is a prompt to build a face sheet. Again, you can take a screenshot of this, hand it to your AI tool, and say, “This is something I’d like to explore. Can you do this for me?” I would start with, “Here is my appointment list,” for whatever your next show is, whether you’re on the meetings side or the leisure side. “These are my appointments. If these things are connected, you can look in my CRM, my email, and my calendar. You can look at my call transcripts and my people and company files, and then you can go to the internet to get information on each person I have an appointment with at this next show.”

“And then I want you to build me a face sheet. I want one card per person.” You saw Sally’s card at the beginning of this call, so you know what those cards can look like. I want it to be a single webpage that I can open on my phone, and it should contain the person’s name, title, company, and photo. Obviously, customize this for whatever you want, but as a starting place: what do you know about that person? Why are you meeting with that person? And give me some ideas for a great opening line. Then give me a way to sort the information, because if you have a hundred appointments, it’s going to be overwhelming to open up this face sheet and see a hundred photos. Am I able to sort by who I met recently, or who I’ve met before at all?

Or whether my company, because I connected this to my CRM, has a relationship with them, or whether this person is new to me. Then you can ask it to populate photos, and I’ll say this is the clunkiest part of the process for me today. If you tell it to limit photos to the company’s website, it will be accurate, but there won’t be a ton of photos. The flip side, and I’ve done this both ways, is to say, “Go to LinkedIn and pull photos from there,” and it will get a lot of wrong images, which is sometimes amusing but not quite as helpful. As you find the correct images for people, you can ask Claude to save that image connected to that person’s Markdown file, so the system gets better and you end up with more images to review over time.

Finally, you want to ask: where did this information come from? This is how I’ve actually used this in the past. If I am going to a dinner and I’ve used Claude to go into Salesforce, I want to know whether Brand USA has a preexisting relationship with the people I’m meeting. If that information is going into my app, I want to see where it came from. I want to see that Salesforce was the source of that information, because that’s how we can have more confidence that nothing was hallucinated. It’s using these connectors to find facts.

These are the places you can look for this type of information. First, the web. It’s going to be very easy for your AI tool to go online and research the people and companies you are meeting with. Every AI tool has internet connectivity at this point, so that’s very simple. It’s the same thing for grabbing a photo if the company’s website has photos of the staff you are meeting with. Then, do you have your own meeting notes? It’s easy enough for your AI tool to look there if you’re using the type of AI tool that has folder access on your computer. It also means you need to have been doing this for a while. If you’re flying out to Vegas over the weekend and you don’t already have all of this set up, there’s not going to be a lot of information for you to pull from.

But it’s quite possible that you can connect your tools to your email. Again, that’s provisioning your IT team needs to set up. But this is a great reason to do it. Getting ready for this call today, I was able to look at the face sheet for this webinar and say, “Oh, I have had email communication with this person or that person,” and it will pull up a reference that jogs my memory: “Oh yeah, I know that person, and we’ve actually interacted before.” Like many of us, I get a lot of inputs into my brain all the time. I need help jogging my memory, particularly before walking into a conference or a dinner where a lot of people will expect that I know them, and that is not my greatest gift.

So every little bit helps, and if it’s a little piece of an email that jogs my memory, I want to have access to that. Then there’s the CRM. If you have the type of CRM that can be connected, that’s a great source of information. But equally great, any of these tools allows you to download a report of who you will be meeting with at a place like IMEX. You can download that report and upload it into the same conversation you are having with your AI tool. In that instance, it’s an extra step compared to it happening automatically, but it’s not a giant extra step. It’s pretty simple, and your AI tool will be able to read that output very easily.

This walks through exporting that report. You want the report to have the contacts at every company you’ll be meeting with and the information you want to include. I’m sure this is work you’ve done before if you work in sales. Excel is fine. CSV is fine. You just attach it to your prompt. And if it’s a short list, maybe you don’t need a report. You can just copy that information and paste it into your AI tool, or take a screenshot of it and paste the screenshot in. There’s no reason for your technology to be a limiting factor in what you’re able to do here. This is accessible at whatever level you’re working.

Then you want to actually check those photos. This is an example from this webinar. It searched 163 different registrants for a photo, and there were twenty that I ended up keeping because I felt confident it was the correct photo and not a wrong one.

This is a demo version that Claude made for me of what the final product looks like. You are able to click on one of those names, and it will bring up additional information about that person. A few months ago, I told Claude I wanted this styled like tarot cards, because I was going to dinners and telling people about this process I’d set up, and people would borrow my phone and hand it around the table. It was like shaking a Magic 8 Ball. They wanted to see what conversation starter Claude had suggested for them. I will say that back when I was doing this, probably in the spring, the information it was giving me on people was a little bit hit and miss.

Part of this is that it gets better over time, because my own notes, that wiki link system, get a little bit richer every single day as I add more information. And the AI tools are getting better as well. A lot of the events I’m using this for are less travel and tourism events and more AI or travel tech events. So it’s people I don’t have any past history with, and what Claude is able to find out about them is limited to what they have written about online. Some of these companies are in stealth and don’t have a lot of information about them online yet. So it’s a little harder to fill this out than if I’m going into something like this webinar today.

Now we’re going back to that original prompt, where we’re looking at Marcus Donnelly and trying to understand what I need to discuss at my meeting with Marcus at IMEX. In this imaginary example, because it is pulling from the CRM, from the web, and from my wiki-linked note-taking system, it is able to instantly pull in richer information about the questions that will likely come up during our meeting.

And I think a really important piece here is to tell AI to let you know what it doesn’t know. Great, you owe this person the request list. We need to find out some other information that this client hasn’t decided on yet. But make sure that when you are instructing Claude to prepare you for these events, you tell it what to do when it doesn’t have information. When people worry about hallucination, this is when hallucinations happen. If a fact is available, it’s not very likely that the AI will make something up rather than use the fact you gave it. Where it will hallucinate is if you ask it whether a request list went out and it doesn’t know the answer. So you have to be explicit and say: for pieces of information where you cannot see an answer, you have to say that you do not know the answer.

Because remember, AI is just desperately trying to make you happy. If it knows that having an answer in that field is what makes you happy, then it’s going to give you an answer. Whereas if it knows that telling you it doesn’t know the answer is what makes you happy, then that is what it will do instead.

For this webinar, I point this out because this is not all perfect yet, because AI doesn’t actually live inside of my brain. I will say Claude put together this entire presentation for me. We have a skill at Brand USA where Claude can create PowerPoint presentations based on an outline that you write with Claude, and then it uses all of our branding to build the deck. When it put together this slide, it said that, based on the face sheet it put together, I know six of the people on the call. But that’s not true. There are another twenty-one people that I do know, but I don’t have any email history with them, and I may not have had a phone call with them since I set up this filing system back in January.

So again, you have to keep the human in the loop, because the AI isn’t going to be magically correct if you haven’t given it a reason to know something. I actually know twenty-seven of the people who registered for the call. A hundred and twenty-eight people on the call are in Salesforce as known contacts to Brand USA, and thirty-three people aren’t connected to Brand USA in Salesforce and don’t have any history with me. We have repeat visitors here, but also a lot of new people, which I think is great. This was all as of this morning. I just went into Claude and said, “Hey, did anybody new sign up for the webinar? What do I need to know going into the presentation today?”

The final step is: great, you have now used Claude to make this face sheet. What do you need to do to put it in your pocket, so it’s actually available to you when you go to a conference? There are a number of ways you can open it. If you’re just on your laptop, you can save the file that ChatGPT or Claude creates for you as an HTML document and open it locally in a browser. Claude will often make HTML pages inside of what it calls an artifact. The artifact is just an interactive way for Claude to communicate with you inside of a chat. If there’s something visual it wants to show you, it will open up a side screen that it calls an artifact, and it will open things like this app inside of that side screen.

If you have the Claude app downloaded on your phone, you’re done. You can go into Claude and open up that face sheet wherever you use Claude. If you truly want it to be a standalone web app, and I have done this before, you can deploy that website to Vercel, which will take support from your technology team. Vercel has password protection available. It’s a small hosting service with free personal accounts. At Brand USA, we have a paid company account, and it’s the paid accounts that are able to put password protection on one of these little websites. That’s a quick way to save it to your home screen so you can pull it up on your phone, and that’s how I was able to pass my phone around the dinner table and let people see what the AI said about them.

I will also say you want to double-check what AI said before you start handing it around to people, because if it thinks it’s writing something for your eyes only, that might not be information you want people to see that you have about them. Don’t let AI put you in an awkward position. And if you are in ChatGPT or Copilot, you can ask for an HTML file, or you can ask for a Word doc as an output that you can then view in OneDrive on your phone.

So that’s how the follow-up you do when you come home from IMEX inherently becomes the prep for your next event, because the system only gets better as more information is added to it. That interconnected web needs lots of connections for the AI tool to see the connections. Right now, your human brain probably sees the connections faster than the AI tool does. Over time, you will see that shift, because as you do this, you will start to be surprised by the connections your AI tool makes, and you’ll think, “Oh, I didn’t realize this person was also part of this other group.”

It starts to bring that level of connection to life for you, and that’s when it starts to become really interesting from a sales perspective.

If you’re looking for next steps: buy the paid account. If your organization does not have a paid AI account yet, we are about to celebrate the fourth anniversary of the release of ChatGPT, so now is the time. A free account trains on whatever your team types into it, and this is information you do not want to put into a free account, because that is how it becomes part of the underlying model the next time OpenAI or Anthropic releases a new version of these tools. That is not how you keep your company’s privileged information safe. You want to use a paid tool so you can turn off things like training on your data for future models. The cost is generally twenty to thirty dollars per person per month, and it really does not matter which tool you use. I personally see Copilot as a few steps behind the other tools.

That is going to change. I just learned on a phone call yesterday that Copilot actually allows you to select between different models that aren’t the Copilot models. One tool might come up with something cool before another one, but they are all reasonably close in capability, and over time, even the weakest ones are only going to get better. So take advantage of whatever your company offers you, but make sure your company is offering a paid tool.

If you are looking for takeaways from this presentation, here’s the slide you can photograph. These are the steps for putting this together yourself: have your AI tool make these folders for you, and set up the prompt where it files your phone call transcripts. You can then start to track how, once you give your AI tool more context, you get better and more useful answers over time. Use this to create a brief. Let’s say you’re not worried about all of the appointments you have next week, just one. Start with that one meeting. What are the priorities of the person you are meeting with? What are three things that came up in the news about their company over the last month?

And what can I say that shows I did my homework to prepare for this meeting? Then, if you are looking to prep for the entire week, build out that face sheet. Share your list of appointments with AI, and give it instructions to make one card per person you have an appointment with, including their name, title, company, a photo if you can source one, what you know about that person, and why you are meeting. And have it say where all of that information was sourced from, so you can be confident it’s true and not hallucinated. That is the invitation I leave you with. Run this on your IMEX meetings if you’re in the meetings space and heading there next week. See how far you can get, and know that the next time you do this with a little more prep time, it’s only going to get better. Start small, but get started.

With that, we have the Q&A open, so I’m watching to see if you have any questions. I invite you to join us for next month’s webinar. The November webinar in the Agents of Change series is going to be about your virtual advisory board: how you can use AI to give you specific feedback, not generic feedback, based not only on thought leaders and their points of view, but maybe on your boss too. Wouldn’t that be helpful? Yes, of course, a recording of this webinar is going to be available. You can find it on our newly relaunched website, thebrandusa.com — thank you, Miles Partnership. That’s our corporate website. We now have a hub you can access from the homepage for AI and innovation, and that is where all of the AI webinars are uploaded, including links to everything I talked about, a transcript, and talking points from the webinars.

It’s a rich resource on how you can use AI at your organization. When you go there, you will also find download instructions for creating your own personal operating system. If you are really interested in pursuing this interwoven, wiki-linked way of working, the piece I walked through today is just one component of that. You can build out the entire system — just steal it from me — and all of the instructions are in the AI and innovation section of our new website. And please connect with me on LinkedIn if we aren’t connected there already.

Now I’m going to go to the Q&A, and we have a question from Sally Berry. Thank you, Sally. “What’s the best way to give AI access to your email and CRM so that it can gather info?” For that, you’ll want to go into the settings of your AI tool. This is the slide that shows you what those settings look like. If you are in Claude or ChatGPT in the desktop app, look in the lower left-hand corner of the app, and you will see your own name or your company name. Click on that, and one of the categories you can navigate to is called Connectors. Connectors will include everything that comes built in to the AI tool, which you can toggle on or off, and you can also add custom connectors there.

Your email, whether you’re using Outlook or Gmail, is going to be on the home screen. Where it says “Your Connectors,” that takes you to a list of all of the available connectors. Email is one of the most popular and common ones, so it’s going to be at the top of that list. Your CRM tool is a separate conversation. You’re only going to see your CRM in the list if it is a huge, very popular tool like Salesforce or HubSpot. If it’s a smaller CRM or one that’s specific to the tourism industry, you will probably need to reach out to your account rep at that company and see if they have a connector available.

They may also call that an MCP connector, or it could be an API connection. MCP stands for Model Context Protocol, and it’s just a way of connecting an AI system to a SaaS tool that is really easy for the AI system to use. It’s possible these companies are working on those connectors. If they are, they’re going to give you a set of instructions for how to set it up. Honestly, I would use your IT team, or if you don’t have an IT team, use your AI tool itself and say, “Hey, here are the instructions I got to set this up. How do I move to the next step? How do I get this actually live?”

We have another question, and this is a great question: “If people keep all of their notes on their local machines as opposed to linking them to a common tool like a CRM, organizations can lose vital information when turnover occurs. Isn’t that an issue?” Yes, absolutely. At the moment, this is not a substitute for keeping your CRM up to date. It wasn’t the topic of this presentation, but I have done other presentations and webinars on how you can use AI to make the process of updating your CRM a little bit easier, because we know that is nobody’s favorite job. In that Salesforce study I showed, I’m sure updating the CRM is part of the work that nobody likes to do, and I don’t blame them.

I don’t like to do that type of work either. So there are a few different solutions. The ideal solution is that at some point it will be simple to have a connector to your CRM that has read and write access. Ideally, I would come home from IMEX, import my transcripts, put together all of my notes, and then just say, “Hey, Claude, would you push all of this information to the correct fields in Salesforce and update Salesforce for me?” That is possible today. That is not a magical future. That is absolutely something AI tools can do today. But I’m going to guess that very few people, if anybody, on this call has a CRM set up where those connections are ready to go.

So there’s an intermediary step, because we’re in this weird interim period where we know what AI can do, but we’re not all able to make use of it yet. I know in Simpleview, for example, you can send them a spreadsheet of information to upload to update all of your appointments. AI can make that spreadsheet for you. You would just give it a blank copy of a template, or examples of similar spreadsheets you have used in the past, or just tell it what fields you want in the spreadsheet. You want to read it over after it builds the spreadsheet, to make sure nothing confidential or personal that wasn’t meant for everybody in the organization ends up in the CRM. But that way, instead of somebody sitting and typing all day to update the CRM one record at a time, it can happen through an upload or some kind of merge tool.

The third way is if there’s no getting around that it’s somebody’s job, maybe your job, to log into the CRM and get everything updated. You can still ask your AI tool to give you all of the information you have to put into the CRM in a format that makes it easy to cut and paste, rather than retyping everything from scratch while looking at your handwritten notes, the way I did when I was working in Broadway ticket sales thirty years ago. And honestly, even if your notes are handwritten, take a photo of them. Your AI tool can turn that photo into typed notes that you can then copy and paste into your CRM.

So this wiki link system I am walking through is meant to supplement your CRM tool. It cannot be a replacement for your CRM tool. But I think we are going to see over time that it all becomes one and the same. Our CRM starts to look like this really valuable, wiki-linked, rich data set that it is everybody’s job to help keep fresh and updated. And I truly believe AI is going to make that so much easier.

We are at the end of the hour, and I don’t see any other questions. So thank you so much to everybody who joined today, and I hope to see you next month for Brand USA’s next Agents of Change webinar. Have a wonderful weekend.